Embedded Character Recognition System using Random Forest Algorithm for IC Inspection System
Character recognition system based on human inspection is unpractical due to lack of accuracy and high cost. Therefore, investigating on automated character inspection system by computer is needed to improve the accuracy, reduce the cost and inspection time. In this project, a Beagle Bone Black (BBB...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
UTeM
2017
|
| Subjects: | |
| Online Access: | http://umpir.ump.edu.my/id/eprint/19731/1/Embedded%20Character%20Recognition%20System%20using%20Random%20Forest%20Algorithm%20for%20IC%20Inspection%20System.pdf http://umpir.ump.edu.my/id/eprint/19731/ http://journal.utem.edu.my/index.php/jtec/article/view/3498 http://journal.utem.edu.my/index.php/jtec/article/view/3498 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Character recognition system based on human inspection is unpractical due to lack of accuracy and high cost. Therefore, investigating on automated character inspection system by computer is needed to improve the accuracy, reduce the cost and inspection time. In this project, a Beagle Bone Black (BBB) was used as a processing device and Logitech webcam was used for as an image acquisition device. Total of 1080 training samples will undergo the image pre-processing, character segmentation, feature extraction and training using random forest classifier. The optimal parameter values of random forest classifier are determined by computing cross validation misclassification rate. The maximum number of splits, number of trees, and learning rate that yields the zero-misclassification rate is 1, 39 and 0.10 respectively. The process of testing random forest classifier was done using SN74LS27N chip under five different illuminations: no LED, one LED, two LED, three LED and four LED. From the experiments, it shows that the proposed system able to achieve 90.00% of accuracy within 1second to recognize characters on the SN74LS27N chip compared to 65.56% accuracy of human inspection. |
|---|